Identification of neurodevelopmental pathway changes in autism spectrum disorder (ASD) datasets
Patients with ASD have deficits in neuronal connectivity as shown by imaging and postmortem studies (McFadden and Minshew 2013). However, the developmental basis of the disruption of CNS connectivity in autism has not been well explored. It is known that multiple developmental processes, including cell differentiation, migration, and proliferation, allow connections to form correctly in the brain. To systematically analyze the developmental deficits in ASD, we retrieved two large datasets: GSE28521, a microarray dataset from the GEO database (Voineagu et al. 2011), and an RNA-seq dataset from the Arkinglab (Gupta et al. 2014). The GSE28521 consists of 32 subjects, including 16 healthy controls and 16 ASD patients. It has transcriptional profiling in the cerebellum (C), frontal cortex (F), and temporal cortex (T) (Fig. 1a, b). After normalization and filtering genes with low expression levels, 10562 genes were detected. Abnormalities have been observed in multiple brain regions in ASD, but unlike the relatively consistent changes in the temporal lobes and frontal lobes, the alterations in the cerebellum are negligible (Riddle et al. 2017)(Turner et al. 2016). This is consistent with the Principal Component Analysis (PCA) result, in which the transcriptional profiling of controls and ASD patients in the cerebellum is indistinguishable (Fig. 1b). Therefore, we only retained the transcriptional profiles from the frontal and temporal cortex for further analysis. PCA also shows that transcriptional profiling in the frontal and temporal cortex shows no differences, and thus for the following analysis, we integrated the data from those locations.
For the datasets in the Arkinglab, 104 subjects including 57 healthy controls and 47 ASD patients were included, encompassing three cortical tissues: Brodmann Area 19 (BA19), Brodmann Area 10 (BA10) and Brodmann Area 44 (BA44) (Gupta et al. 2014). After normalization, 19371 genes were detected (Fig. 1c, d).
To understand the developmental processes, 5 development-related gene sets were examined: synapse assembly (GO: GO:0007416), neuroblast proliferation (GO:0007405), regulation of neural precursor cell proliferation (GO:2000177), axon guidance (GO: 0007411), and cell differentiation (GO: 0030154). Differentially expressed gene sets were analyzed by the ROAST test (Wu et al. 2010).
Expression changes in most gene sets were consistent in these two datasets. For instance, the gene sets labeled synapse assembly (p = 0.001, 0.012, respectively), neuroblast proliferation (p = 0.001, 0.034, respectively) and axon guidance (p = 0.001, 0.006, respectively) were significantly upregulated, while regulation of neural precursor cell proliferation (p = 0.004, 0.059, respectively) was downregulated in both datasets. Cell differentiation (p = 0.001, 0.002, respectively) was the only gene set with an opposite change in direction between the two datasets(Fig. 2a).
Identification of changes in dopamine-related gene sets in ASD
Altered dopaminergic signaling, such as changes in dopamine-hydroxylase activity (Garnier 1986)(Lake and Murphy 1977) and dopamine levels (Launay et al. 1987), has been observed in the blood and urine of ASD patients (Clin 1992). However, a systematic analysis of dopaminergic signaling changes in the brain has not been done.
To explore this, multiple dopamine-related gene sets were selected for further analysis: the dopamine catabolic process (GO:0042420), dopamine receptor activity (GO:0004952), dopamine receptor binding (GO:0050780), dopamine secretion (GO:0014046), dopamine transport (GO:0015872), dopamine uptake (GO:0090494), L-dopa decarboxylase activity (GO:0036468), regulation of the dopamine biosynthetic process (GO:1903179), and regulation of the dopamine receptor signaling pathway (GO:0060159). All of these pathways or genes sets can regulate dopaminergic signaling.
Excepting dopamine transport and L-dopa decarboxylase, the expression of most dopamine-related gene sets was altered. Dopamine receptor activity (p = 0.001, 0.022, respectively), dopamine secretion (p = 0.001, 0.001, respectively), regulation of dopamine receptor signaling pathway (p = 0.001, 0.001, respectively), and dopamine uptake ( p = 0.001, 0.004, respectively) were upregulated, while the dopamine catabolic process ( p = 0.041, 0.012, respectively) was downregulated in both Arkinglab and GSE28521 datasets(Fig. 2b). Discrepancies in gene set expression changes were also observed. The direction of the dopamine receptor binding gene set was the opposite in the Arkinglab and the GSE28521 data sets (p = 0.001, 0.004, respectively). The dopamine biosynthetic process gene set was upregulated in the Arkinglab data set, but did not show differential expression in the GSE28521 data set (Table 1, p = 0.001, 0.348, respectively).
Table 1
Expression changes of development-related gene sets and dopaminergic gene sets in the GSE28521 and Arkinglab datasets.
|
Arkinglab
|
GSE28521
|
|
Direction
|
PValue
|
PValue.Mixed
|
Direction
|
PValue
|
PValue.Mixed
|
synapse assembly
|
Up
|
0.001
|
0.001
|
Up
|
0.08
|
0.012
|
neuroblast proliferation
|
Up
|
0.073
|
0.001
|
Up
|
0.131
|
0.034
|
regulation of neural precursor cell proliferation
|
Down
|
0.006
|
0.004
|
Down
|
0.053
|
0.059
|
axon guidance
|
Up
|
0.003
|
0.001
|
Up
|
0.44
|
0.006
|
cell differentiation
|
Up
|
0.591
|
0.001
|
Down
|
0.002
|
0.002
|
dopamine catabolic process
|
Down
|
0.016
|
0.041
|
Down
|
0.1
|
0.012
|
dopamine receptor activity
|
Up
|
0.001
|
0.001
|
Up
|
0.022
|
0.022
|
dopamine receptor binding
|
Up
|
0.321
|
0.001
|
Down
|
0.029
|
0.004
|
dopamine secretion
|
Up
|
0.001
|
0.001
|
Up
|
0.002
|
0.001
|
dopamine transport
|
Up
|
0.232
|
0.069
|
NA
|
NA
|
NA
|
dopamine uptake
|
Up
|
0.006
|
0.001
|
Up
|
0.004
|
0.016
|
L-dopa decarboxylase activity
|
Down
|
0.689
|
0.689
|
Up
|
0.06
|
0.06
|
regulation of dopamine biosynthetic process
|
Down
|
0.293
|
0.315
|
Up
|
0.008
|
0.008
|
regulation of dopamine receptor signaling pathway
|
Up
|
0.001
|
0.001
|
Up
|
0.038
|
0.001
|
dopamine biosynthetic process
|
Up
|
0.159
|
0.001
|
Up
|
0.242
|
0.348
|
Correlation between dopamine- and neurodevelopment-related gene sets
In the past, dopamine’s role has almost entirely been relegated to that of a neurotransmitter, regulating a series of physical processes (Basu and Sarathi Dasgupta 2000)(Oishi and Lazarus 2017)(Chen et al. 2017). Recent evidence shows that dopaminergic signaling is also important for neural development (Cai et al. 2021). However, the contribution of dopaminergic signaling to the neurodevelopmental deficits observed in ASD has not been explored. To analyze the relationship between dopamine- and neurodevelopment-related gene sets, Gene Set Causal Relationship Analysis was employed, which reveals the expression correlation between gene sets regardless of their direct or indirect interaction (Yue et al. 2018).
In the dataset GSE28521, we found that the cell differentiation gene set was highly correlated with most dopamine-related gene sets, in particular with dopamine receptor binding. Neuroblast proliferation was also correlated with the dopamine receptor binding gene set. The axon guidance gene set was correlated with the gene sets of dopamine receptor binding, dopamine secretion, and the dopamine biosynthetic process. The synapse assembly gene set was positively correlated to the dopamine receptor binding gene set and the dopamine biosynthetic process gene set (Fig. 3a, Table 2). In the Arkinglab datasets, similar results were observed, for example, cell differentiation was correlated to many dopamine-related gene sets.
Table 2
Correlations between development-related gene sets and dopaminergic gene sets in the GSE28521 dataset.
Development-related gene sets
|
Dopaminergic gene sets
|
Direction
|
P value
|
axon guidance
|
dopamine biosynthetic process
|
positive
|
0.00220458
|
axon guidance
|
dopamine receptor binding
|
positive
|
2.79E-09
|
axon guidance
|
dopamine secretion
|
positive
|
1.78E-13
|
synapse assembly
|
dopamine biosynthetic process
|
positive
|
0.02159671
|
synapse assembly
|
dopamine receptor binding
|
positive
|
0.01058364
|
cell differentiation
|
dopamine biosynthetic process
|
negative
|
2.40E-14
|
cell differentiation
|
dopamine receptor binding
|
negative
|
1.60E-50
|
cell differentiation
|
dopamine secretion
|
negative
|
4.58E-37
|
cell differentiation
|
dopamine uptake
|
negative
|
9.73E-18
|
cell differentiation
|
regulation of dopamine receptor signaling pathway
|
negative
|
1.06E-07
|
neuroblast proliferation
|
dopamine receptor binding
|
negative
|
0.003305
|
Clinical symptoms in ASD are heterogeneous, and age is an important factor to consider (Ewen et al. 2019). To test the correlation between dopamine- and development-related gene sets, we selected young subjects (age ≤ 18) in the Arkinglab dataset for further analysis. Among these younger subjects, the cell differentiation gene sets were positively correlated with the dopamine receptor binding and dopamine secretion gene sets. The axon guidance gene set was negatively correlated with dopamine uptake gene set, while positively correlated with the dopamine receptor binding gene set. The dopamine biosynthetic process gene set was negatively correlated with the axon guidance gene set. Dopamine receptor binding and dopamine secretion were positively correlated to synapse assembly, but dopamine uptake was negatively correlated with synapse assembly (Fig. 3b, Table 3). When only the subjects with rapid brain development (age ≤ 10) were selected, the results were consistent except that the dopamine uptake gene set was negatively correlated with the cell differentiation gene set (Fig. 3c, Table 4).
Table 3
Correlations between development-related gene sets and dopaminergic gene sets in the Arkinglab dataset (age < 18).
Development-related gene sets
|
Dopaminergic gene sets
|
Direction
|
Pvalue
|
axon guidance
|
dopamine receptor binding
|
positive
|
5.78E-63
|
axon guidance
|
dopamine biosynthetic process
|
negative
|
8.18E-05
|
axon guidance
|
dopamine uptake
|
negative
|
1.29E-16
|
cell differentiation
|
dopamine secretion
|
positive
|
5.61E-76
|
synapse assembly
|
dopamine receptor binding
|
positive
|
3.11E-09
|
synapse assembly
|
dopamine secretion
|
positive
|
0.011703
|
synapse assembly
|
dopamine uptake
|
negative
|
0.0010405
|
Table 4
Correlations between development-related gene sets and dopaminergic gene sets in the Arkinglab dataset (age < 10).
Development-related gene sets
|
Dopaminergic gene sets
|
Direction
|
Pvalue
|
axon guidance
|
dopamine receptor binding
|
positive
|
7.16E-57
|
axon guidance
|
dopamine uptake
|
negative
|
6.06E-15
|
cell differentiation
|
dopamine receptor binding
|
positive
|
3.69E-141
|
cell differentiation
|
dopamine secretion
|
positive
|
1.18E-60
|
cell differentiation
|
dopamine uptake
|
negative
|
0.0096915
|
synapse assembly
|
dopamine receptor binding
|
positive
|
9.51E-09
|
synapse assembly
|
dopamine secretion
|
positive
|
0.0450494
|
synapse assembly
|
dopamine uptake
|
negative
|
0.0018291
|
The effects of dopaminergic signaling on neurodevelopment
To test the potential role of dopamine in neural circuit formation in vivo, we used zebrafish as a model. As a vertebrate, zebrafish has been widely used in the study of neurodevelopment and neurological disorders(Chakravarty et al. 2013). Non-lethal concentrations (0.2 µM to 1.5 µM) of the dopamine antagonist SCH 23390 were administrated from 1 dpf (day post-fertilization) to 5 dpf when neural circuits were rapidly generated (including the period of neurogenesis and increases in CNS connectivity). We did not see significant morphological changes in brain and body development (Fig. 4a-d).
To determine the role of dopamine in neural circuits, we used the smallest dosage of SCH 23390 (0.2 µM). We found that number of neurons labeled by the red fluorescent protein mCherry in the transgenic line Tg(elavl3:mCherry) (elavl3 is a pan-neuronal marker) was decreased in the telencephalon and diencephalon (Fig. 5c, p = 0.0003, n = 7 and 7, Fig. 5f,p < 0.0001,n = 7 and 7, respectively), corresponding to the forebrain and midbrain in mammals, respectively. This was consistent with the defects observed in autistic patients (Orefice et al. 2016). When we used tubulin as a marker for mature neuronal and axon connectivity, we found a dramatic decrease in staining when larvae were treated with the drug SCH 23390, especially in the telencephalon (Fig. 5i, p = 0.0022, n = 6 and 7, respectively).
We also used staining for SV2 (Synaptic Vesicle protein 2, another mature neuron or presynaptic marker) to determine the effects of SCH 23390 on neural circuits. Consistently, SV2 staining was dramatically decreased (Fig. 6A, B, p = 0.0005, n = 10 and 11, respectively). To confirm the specific role of dopamine, we generated an otp mutant in which the otpa and otpb genes were disrupted (Fig. 6d, e). Otp is a transcription factor critical for dopamine synthesis(Ding et al. 2020). Synapses labeled by SV2 were frequently disrupted in the otp mutants (Fig. 6a, c, p = 0.0027, n = 10 and 9, respectively), indicating that dopaminergic signaling can affect the maturation of neural circuits.
Systematic analysis of downstream pathways of dopaminergic signaling
To further test potential downstream pathways affected by dopaminergic signaling during development, we dissected zebrafish brains after treatment with SCH23390 at 5 days post-fertilization. We found 2401 DEGs (differentially expressed genes) compared to controls (Fig. 7a). Interestingly, most DEGs were enriched in the pathways related to cell adhesion, including: homophilic cell adhesion via plasma membrane adhesion molecules and cell-cell adhesion via plasma-membrane adhesion molecules (Fig. 7b). These pathways are critical for multiple neural developmental processes and neural circuit formation (Peglion et al. 2014).
Transcription factors among the genes were uploaded for protein-protein interaction (PPI) analysis with STRING (The search tool for retrieval of interacting genes, https: //string-db. org). All together, we found significant interactions among these transcription factors as shown in the PPI network (Fig. 7d). Transcription factors including fos, gata4 and sta1b have the strongest connections with other proteins. These proteins were reported to have multiple roles in neurodevelopment, including neurogenesis and axon guidance(Cao et al. 2020)(Duan et al. 2018). Altogether, this evidence suggests a role for dopamine in neural circuit formation and in pathways important in neurodevelopment.